Main

Chronic inflammation in times of nutritional excess has been proposed to be critical in regulating the development of obesity and metabolic syndrome7,8; however, the molecular factors that are involved have not been well defined. In an attempt to screen the inflammation-associated factors that are related to obesity, we determined the serum levels of different cytokines and found that the IL-12 family members (IL-27p28, IL-35 and IL-12p40) were significantly decreased in human participants with obesity (Extended Data Fig. 1a and Supplementary Table 1). Genome-wide association studies revealed that single-nucleotide polymorphisms in IL27 (which encodes IL-27p28) are associated with body mass index (BMI)9 and insulin resistance10. We therefore focused on the potential regulation of obesity by IL-27 signalling. IL-27 is a heterodimer composed of Epstein–Barr virus induced 3 (EBI-3) and p28, and regulates immune responses with pleiotropic effects11. We next examined the serum level of intact IL-27 using an enzyme-linked immunosorbent assay (ELISA) and found that it was indeed decreased in the group with obesity (Fig. 1a). Furthermore, the level of IL-27 further decreased as the degree of obesity increased (Extended Data Fig. 1b), and it was also inversely correlated with the fasting glucose levels of individuals with type 2 diabetes (Extended Data Fig. 1c and Supplementary Table 2). Interestingly, the IL-27 level of individuals with obesity was restored after bariatric surgery that reduced their body weight (Fig. 1b and Supplementary Table 3). These results strongly suggest that IL-27 signalling has a role in obesity and type 2 diabetes.

Fig. 1: IL-27 restrains the development of obesity and insulin resistance.
figure 1

a, The serum levels of IL-27 in individuals with obesity (Ob) (n = 32) and control individuals with a healthy BMI (lean) (n = 30). b, The BMI and serum IL-27 level of patients with obesity before (pre) and one month after (post) gastric bypass surgery. n = 7. c–g, Il27ra-KO mice and WT controls (aged 8 weeks) were treated with an HFD or ND for 10 weeks. c, Body weight was recorded each week. n = 5 (WT, ND), n = 6 (WT, HFD; and KO, ND) and n = 7 (KO, HFD). d, Tissues were collected and weighed. n = 10 (WT, SCW and liver), n = 9 (KO, SCW and liver) n = 11 (WT, epididymal white adipose tissue (eWAT)) and n = 12 (KO, eWAT). e, f, Intraperitoneal glucose tolerance test (GTT) (e; n = 6 (ND), n = 5 (KO, HFD) and n = 4 (WT, HFD)) and insulin tolerance test (ITT) (f; n = 5 (WT, ND), n = 6 (WT, HFD; and KO, ND) and n = 7 (KO, HFD). g, HFD-fed WT and Il27ra-KO mice (10 weeks) were tested for insulin sensitivity through immunoblot analysis of phosphorylated AKT (p-AKT S473) in epididymal fat. h–j, Ebi3-KO mice and WT controls (aged 8 weeks) were treated with an HFD for 9 weeks. h, Body weights were recorded each week. n = 6 (WT) and n = 5 (KO). i, j, GTT (i) and ITT (j) analyses were performed. n = 8 (WT) and n = 4 (KO). Data are mean ± s.e.m. of biologically independent samples. Statistical analysis was performed using two-tailed unpaired Student’s t-tests (a and d), two-tailed paired Student’s t-tests (b), two-way analysis of variance (ANOVA) (c and h) and two-way ANOVA with Sidak’s multiple-comparisons test (e–f and i–j).

Source data

Full size image

To investigate the role of IL-27 signalling in the development of obesity, we first fed IL-27Rα-deficient mice (Il27ra-knockout (KO) (also known as Wsx1-KO) mice) and C57BL/6J wild-type (WT) control mice with a high-fat diet (HFD). Il27ra-KO mice were highly susceptible to HFD-induced obesity (Fig. 1c, d and Extended Data Fig. 2a) and developed more severe glucose intolerance, insulin resistance (Fig. 1e–g) as well as steatohepatitis (Extended Data Fig. 2b, c). The serum levels of leptin and adiponectin in Il27ra-KO mice were increased (Extended Data Fig. 2d). Consistent with these metabolic phenotypes, HFD-fed Il27ra-KO mice also developed enhanced inflammatory responses in adipose tissue, as indicated by increased macrophage infiltration and higher inflammatory cytokine production (Extended Data Fig. 2e, f). As Il27ra-KO mice have an approximately C57BL6/NJ genotype, we also used C57BL/6N WT mice as the control group and observed similar phenotypes (Extended Data Fig. 2g–j). Even Il27ra+/− heterozygote mice showed considerably increased body weight and impaired insulin sensitivity after feeding on an HFD, suggesting a dose-dependent effect of IL-27 signalling (Extended Data Fig. 2k–n). Interestingly, glucose metabolism was already impaired in Il27ra-KO mice after 4 weeks of HFD feeding (Extended Data Fig. 2o), despite the fact that no significant difference in body weight was observed compared with the WT controls at this time (Fig. 1c). This suggests that there was a precommitted metabolic alteration after IL-27 signalling deficiency. Moreover, we also found that mice that were deficient in EBI-3 were more susceptible to HFD-induced obesity and glucose intolerance (Fig. 1h–j). Collectively, these results strongly demonstrate a protective role of IL-27 signalling in restraining obesity and related metabolic morbidities.

We next examined whether IL-27 signalling deficiency affects energy balance. Il27ra-KO mice that were maintained on a normal diet (ND) showed significantly reduced energy expenditure without altered food intake (Extended Data Fig. 3a–c). The phenotype was even more striking when they were fed an HFD (Fig. 2a–c). An analysis of adipose tissues revealed reduced oxygen consumption (Extended Data Fig. 3d), indicating a promoting action of IL-27 signalling in maintaining energy homeostasis. To determine the molecular mechanisms that underlie the impairment of energy expenditure after IL-27Rα deficiency, we performed a transcriptomic analysis of subcutaneous white adipose tissue (SCW) from HFD-fed Il27ra-KO or WT mice. Gene set enrichment analysis revealed that IL-27Rα deficiency resulted in alterations in multiple cellular functions and signalling pathways (Supplementary Table 4), including significant impairment of the brown/beige adipocyte thermogenesis program and PPAR signalling pathway (Extended Data Fig. 3e, f). A reduced expression of several thermogenic genes in Il27ra-KO mice was detected even in mice that were fed a ND (Extended Data Fig. 3g).

Fig. 2: IL-27 signalling promotes thermogenesis and energy expenditure.
figure 2

a–c, Il27ra-KO and WT mice were fed an HFD for 4 weeks and then placed into metabolic cages. Food intake (a; n = 7), oxygen consumption (\({V}_{{{\rm{O}}}_{2}}\); b; n = 7 (WT) and n = 8 (KO)) and energy expenditure (c; n = 7 (WT) and n = 8 (KO)) were monitored over a 24 h period. d, e, Il27ra-KO and WT mice were fed an HFD for 6 weeks and then cold challenged at 4 °C. The survival curve (d; n = 8 (WT) and n = 6 (KO)) and rectal temperature (e; n = 15 (WT) and n = 11 (KO)) are shown. f, g, Il27ra-KO and WT mice were fed an HFD for 6 weeks and then housed at 25 °C (f) or challenged at 4 °C for 2 h (g). SCW and BAT samples were collected for immunoblot analysis of UCP1. h–j, Il27ra-KO and WT mice fed a ND were housed at 25 °C or cold challenged at 4 °C for 48 h. h, SCW and BAT samples were collected for immunoblot analysis of UCP1. i, Immunohistochemical staining of UCP1. Scale bars, 100 µm. j, Haematoxylin and eosin (H&E) staining of BAT and SCW tissues. Scale bars, 50 µm. Data are mean ± s.e.m. of biologically independent samples. Statistical analysis was performed using two-way ANOVA (a–c), two-way ANOVA with Sidak’s multiple-comparisons test (e) and log-rank tests (d).

Source data

Full size image

Brown/beige adipocyte thermogenesis has a key role in driving energy expenditure that has increasingly been considered to be a target for combating the development of obesity and metabolic syndrome1,2. To determine whether the effect of IL-27Rα deficiency on energy expenditure is attributable to decreased heat production, we next analysed adaptive thermogenesis. Consistent with their increased susceptibility to diet-induced obesity, Il27ra-KO mice were hypersensitive to cold-induced hypothermia when fed the ND (Extended Data Fig. 3h, i), and this cold intolerance was even more severe after HFD feeding (Fig. 2d, e). Consistently, expression of uncoupling protein 1 (UCP1), the key mitochondrial protein in thermogenesis, was also markedly reduced in the SCW of Il27ra-KO mice that were fed the HFD or ND, either after being maintained at 25 °C or after cold exposure at 4 °C (Fig. 2f–i). Histological analyses also showed less cells with multilocular lipid droplets (Fig. 2j) in the SCW of Il27ra-KO mice. The reduction of UCP1 in the brown adipose tissues (BAT) was either less substantial (Fig. 2h) or largely unchanged (Fig. 2f, g), suggesting that the effect of IL-27 signalling on the thermogenesis program occurred mainly through actions in the SCW, and presumably IL-27 regulates beige adipocytes more robustly than brown adipocytes. Note that UCP1 was already decreased in Il27ra-KO mice without cold stimulation (Fig. 2f, h), which was in accordance with the decreased systemic energy expenditure (Fig. 2b, c and Extended Data Fig. 3b, c). Moreover, a cold challenge test in Ebi3-KO mice recapitulated the findings in Il27ra-KO mice (Extended Data Fig. 3j–l), and supplementing IL-27 in Ebi3-KO mice upregulated UCP1 and ameliorated the cold-induced hypothermia (Extended Data Fig. 3m–p), further verifying the thermogenic effects of IL-27 signalling.

Previous studies showed that immune cells, especially lymphoid lineage cells, express high levels of IL-27R11 and the immune system was known to be involved in the development of obesity7. We therefore next investigated whether IL-27 regulates the development of obesity through actions on these immune cells. We generated Il27raflox/flox mice (Extended Data Fig. 4a–c) and crossed them with Cd2-cre or Lyz2-cre mice to specifically delete Il27ra in lymphoid cells (T cells and B cells) or myeloid cells (monocytes, mature macrophages and granulocytes), respectively (Extended Data Fig. 4d). Surprisingly, deficiency of IL-27Rα in these cells did not alter their susceptibility to HFD-induced obesity or the related metabolic syndrome (Extended Data Fig. 4e–h). To determine which cell types were responsible for IL-27-mediated signalling in obesity, we generated four groups of chimera mice (bone marrow from WT mice in WT mice (WT > WT), WT > KO, KO > WT and KO > KO) using bone marrow from Il27ra-KO or WT mice and fed these mice an HFD. Interestingly, only those chimeras that were prepared using Il27ra-KO recipient mice regardless of the donor cells were highly susceptible to HFD-induced obesity and metabolic syndrome (Fig. 3a and Extended Data Fig. 5a–d). Consistently, the thermogenesis of WT > KO chimeras was dampened compared with WT > WT chimeras as indicated by lower hypothermia, less multilocular lipid droplets and reduced expression of UCP1 in the SCW (Fig. 3b, c and Extended Data Fig. 5e–g). These results strongly suggest that the dominant targets of IL-27 signalling in mediating metabolic homeostasis were not radiation-sensitive immune cells. Direct actions of IL-27 on non-immune cells were reported recently12 and Il27ra transcripts were also detected in adipocytes13. Indeed, IL-27Rα protein was detectable in the adipocyte fraction of adipose tissues as well as in in vitro-differentiated primary beige adipocytes (Extended Data Fig. 6a, b). The expression of IL-27Rα in the SCW was much higher than that in the BAT, which coincided with more striking phenotypes in the SCW. To test whether IL-27 could directly target adipocytes to regulate thermogenesis and obesity, we crossed Il27raflox/flox mice with Adipoq-cre mice to specifically delete Il27ra in adipocytes (Extended Data Fig. 6c, d). Abolishing IL-27Rα signalling in adipocytes predisposed the mice to hypersensitivity to HFD-induced obesity and the related metabolic syndrome (Fig. 3d and Extended Data Fig. 6e–k), which recapitulated our findings in mice with a global KO of Il27ra. Furthermore, the thermogenesis of Adipoq-cre;Il27raflox/flox mice was also dampened as indicated by intolerance to cold challenge, decreased UCP1 expression and less multilocular lipid droplets in the SCW (Fig. 3e, f and Extended Data Fig. 6l, m). As Adipoq-cre;Il27raflox/flox deletes Il27ra in all fat cells, we also examined whether IL-27Rα signalling in beige/brown adipocytes could regulate thermogenesis by crossing Il27raflox/flox with Ucp1-creERT2 mice to specifically delete Il27ra in beige/brown adipocytes after tamoxifen induction. Given that UCP1 is expressed in most adipocytes in early adipose tissue development14, we used this inducible conditional KO strain rather than Ucp1-cre to avoid possible developmental effects. Ucp1-creERT2;Il27raflox/flox mice were also more sensitive to HFD-induced obesity (Fig. 3g and Extended Data Fig. 7a–c) and cold-induced hypothermia (Fig. 3h, i and Extended Data Fig. 7d) compared with tamoxifen-treated Il27raflox/flox control mice, although the differences were less substantial compared with in Adipoq-cre;Il27raflox/flox mice (Fig. 3d and Extended Data Fig. 6f, g) or mice with a global KO of Il27ra (Fig. 1c–f). The diminished phenotypes in Ucp1-creERT2;Il27raflox/flox mice might be due to the reduced but not complete loss of IL-27Rα in the SCW adipocyte fraction (as not all adipocytes express UCP1; Extended Data Fig. 7a), which suggests that expression of IL-27Rα in UCP1low or UCP1− adipocytes also has an important role in these processes. Collectively, these data clearly demonstrate that IL-27 directly targets adipocytes to promote thermogenesis and protect against obesity, with beige/brown adipocytes serving as important responders.

Fig. 3: IL-27 directly targets adipocytes to promote thermogenesis and combat obesity.
figure 3

a, Four groups of bone marrow chimeras were generated using Il27ra-KO or WT mice and fed an HFD for 10 weeks. The body weights were recorded weekly. n = 14 (WT > WT), n = 11 (WT > KO), n = 12 (KO > WT), n = 10 (KO > KO). NS, not significant. b, The rectal temperature of chimeric mice fed normal chow in response to a cold challenge at 4 °C. n = 7 (WT > WT) and n = 6 (WT > KO). c, After the cold challenge for 12 h, SCW samples were collected for immunoblot analysis. d, Adipoq-cre;Il27raflox/flox (n = 7) and Il27raflox/flox (n = 9) mice (aged 8 weeks) were fed an HFD for 10 weeks. Body weight was recorded weekly. e, The rectal temperature of mice fed an ND in response to cold challenge at 4 °C. n = 16 (Il27raflox/flox) and n = 17 (Adipoq-cre;Il27raflox/flox). f, After the cold challenge for 12 h, SCW and BAT samples were collected for immunoblot analysis. g, Ucp1-creERT2;Il27raflox/flox (n = 12) and Il27raflox/flox (n = 11) mice (aged 8 weeks) were pretreated with tamoxifen and then fed an HFD for 10 weeks. Body weight was recorded weekly. h, The rectal temperature of mice fed an ND in response to cold challenge at 4 °C. n = 4 (Il27raflox/flox) and n = 6 (Ucp1-creERT2;Il27raflox/flox). i, After the cold challenge for 12 h, SCW and BAT samples were collected for immunoblot analysis. Data are mean ± s.e.m. of biologically independent samples. Statistical analysis was performed using two-way ANOVA (a and d) and two-way ANOVA with Sidak’s multiple-comparisons test (b, e, g and h).

Source data

Full size image

To dissect the molecular mechanism by which IL-27 regulates thermogenic adipocytes, we directly treated primary beige adipocytes with IL-27 in vitro. The expression of UCP1 was substantially increased after IL-27 treatment, together with upregulated peroxisome proliferator-activated receptor alpha (PPARα) and peroxisome proliferator-activated receptor gamma co-activator 1alpha (PGC-1α), the two key transcriptional activators that govern the energy metabolism15,16 (Fig. 4a and Extended Data Fig. 7e). The canonical downstream signalling pathways of IL-27Rα are mediated by STAT1, STAT3 and p38 MAPK in immune cells11,17. We next determined whether IL-27 promotes the activation of thermogenesis in adipocytes through these pathways. The phosphorylation of p38 MAPK was significantly increased after IL-27 stimulation, and the phosphorylation activation of STAT3 was only transiently induced and rapidly faded, while the phosphorylation of STAT1 was barely detectable (Fig. 4b and Extended Data Fig. 7f). It is well established that p38 MAPK instigates ATF2-mediated upregulation of PGC-1α—the master regulator of UCP1 and thermogenesis15,16,18,19. Indeed, IL-27 treatment augmented ATF2 activation in primary beige adipocytes with ensuing upregulation of PGC-1α, which could be abrogated by pharmacological inhibition of the activity of p38 MAPK (Fig. 4a–c). Moreover, cells with p38 MAPK inhibition were refractory to IL-27 induced promotion of UCP1 expression (Fig. 4c and Extended Data Fig. 7g, h). However, IL-27 upregulation of UCP1 was largely unaffected in response to STAT3 inhibition, even though it also decreased PGC-1α expression (Fig. 4c and Extended Data Fig. 7g, h). These phenomena disappeared when the IL-27 treatment was performed in Il27ra-KO mouse primary beige cells (Fig. 4c). These results indicate that IL-27 could promote the activation of thermogenesis mainly through the p38 MAPK–PGC-1α signalling pathway. Interestingly, neither p38 MAPK nor STAT3 inhibition abrogated the upregulation of PPARα by IL-27 (Fig. 4c (top)), suggesting additional pathways might be involved in mediating IL-27 actions. An unexpected upregulation of PGC-1α was detected after treatment with the STAT3 inhibitor or p38 inhibitor alone in the absence of IL-27 stimulation, and STAT3 inhibitor alone appeared to slightly increase UCP1 expression (Fig. 4c and Extended Data Fig. 7h), suggesting that STAT3 and p38 MAPK might have complicated actions in thermogenesis under unstimulated conditions.

Fig. 4: IL-27 promotes the activation of thermogenesis with promising therapeutic potential.
figure 4

a, Primary beige adipocytes were generated in vitro from the stromal vascular fraction of WT SCW, treated with rmIL-27 (100 ng ml−1) or PBS for 24 h and used for immunoblotting analysis. b, Immunoblotting analysis of protein phosphorylation of WT primary beige adipocytes in response to treatment with rmIL-27 (100 ng ml−1). c, Primary beige adipocytes from WT and Il27ra-KO SCW samples were generated in vitro. Cells were treated with STAT3 inhibitor (STAT3i; C188-9, 10 µM) or p38 MAPK inhibitor (p38i; SB203580, 10 µM) and used for immunoblotting analysis of the interested proteins. d–i, WT mice were fed an HFD for 32 weeks and then intraperitoneally injected with rmIL-27 (100 µg kg−1) or PBS every other day for 15 d. d, Body weight was recorded at indicated time points. n = 8. e–g, Tissues were weighed (e; n = 6 (SCW), n = 8 (eWAT), n = 7 (liver, PBS) and n = 8 (liver, IL-27)), and GTT (f; n = 12 (PBS) and n = 11 (IL-27)) and ITT (g; n = 9 (PBS) and n = 8 (IL-27)) measurements were performed, after 15 d treatment. h, Immunoblotting analysis of phosphorylated AKT (p-AKT S473) in eWAT was performed for insulin sensitivity analysis and shown. i, Oil red O staining of liver tissues after rmIL-27 therapy for 15 d. Scale bars, 50 µm. Data are mean ± s.e.m. of biologically independent samples. Statistical analysis was performed using two-tailed unpaired student’s t-tests (e), two-way-ANOVA (d) and two-way ANOVA with Sidak’s multiple-comparisons test (f and g).

Source data

Full size image

As IL-27 signalling promotes thermogenesis and restrains obesity, and individuals with obesity showed significantly decreased serum IL-27 levels (Fig. 1a), we next sought to test whether IL-27 has therapeutic potential. We treated obese WT mice with IL-27 and observed significantly reduced body weight and adipose deposition (Fig. 4d, e). More importantly, IL-27 administration also substantially ameliorated insulin resistance and hepatic steatosis (Fig. 4f–i). These findings revealed a strong therapeutic potential of triggering the IL-27 signalling pathway for the treatment of obesity and insulin resistance. We also applied IL-27 therapy to obese Il27raflox/flox, Adipoq-cre;Il27raflox/flox, Ucp1-creERT2;Il27raflox/flox and Ucp1-KO mice. IL-27 injection did not cause systemic inflammation or tissue damage (Extended Data Fig. 8a–c). Consistent with the findings in WT mice, IL-27 administration also significantly reduced the body weight and improved glucose metabolism in Il27raflox/flox control mice (Extended Data Fig. 8d–f). By contrast, theprotective effects of IL-27 therapy vanished in Adipoq-cre;Il27raflox/flox and Ucp1-KO mice, and diminished in Ucp1-creERT2;Il27raflox/flox mice (Extended Data Fig. 8g–p). These data indicate that the beneficial effects of IL-27 treatment depend on its actions on adipocytes, with beige/brown adipocytes as the key responders. The protective effects of IL-27 therapy in Ucp1-creERT2;Il27raflox/flox mice were not completely abolished as in Adipoq-cre;Il27raflox/flox mice (Extended Data Fig. 8g–l), suggesting again that IL-27Rα signalling in additional types of adipocytes might also have an important role in these processes. The loss of the therapeutic effects of IL-27 in obese Ucp1-KO mice (Extended Data Fig. 8m–p) suggested that UCP1 is required for the metabolic improvement by IL-27.

Myeloid cells such as macrophages and dendritic cells are important IL-27 producers11, and the expression of IL-27 in adipocytes was also reported previously13. We next investigated whether the IL-27 was produced from these cells to restrain obesity. We crossed Il27p28flox/flox mice with Lyz2-cre, Itgax-cre or Adipoq-cre mice to specifically eliminate the production of IL-27 from monocytes, mature macrophages and granulocytes; dendritic cells and M1 macrophages; or adipocytes, respectively, and challenged these mice with an HFD. Unexpectedly, none of these mice recapitulated the findings in IL-27Rα-deficient or Ebi3-KO mice (Extended Data Fig. 9a–i). We next sought to test whether the source of IL-27 in the adipose tissue might be CX3CR1+ cells, which contain sympathetic-neuron-associated macrophages that regulate thermogenesis and accumulate in obesity20. Cx3cr1-cre;Il27p28flox/flox mice showed a tendency to gain more weight after feeding with an HFD (Extended Data Fig. 9j). More importantly, these mice displayed worse glucose intolerance and insulin resistance after 10 weeks of HFD feeding (Extended Data Fig. 9k, l), indicating that CX3CR1+ cells may represent an important source of IL-27 in combating obesity. As the CX3CR1+ cohorts are very complicated21, the precise cell types of these CX3CR1+ cells remain to be delineated. Besides, the increase of body weight after IL-27p28 deletion in CX3CR1+ cells was much less substantial than that in IL-27Rα- or EBI-3-deficient mice (Extended Data Fig. 9j), suggesting that there might be other sources of IL-27 that also contribute to obesity regulation. Inflammatory cytokines such as IFNs and TLR ligands such as LPS are known to stimulate the production of IL-27 (ref. 11), but whether these signals are involved in affecting the production of IL-27 or IL-27-producing cells during obesity-related metabolic inflammation needs further investigation.

In summary, our studies have uncovered that IL-27 signalling is required for intact adaptive thermogenesis and systemic metabolic homeostasis. IL-27 directly targets adipocytes to elicit the activation of p38 MAPK, thereby driving the activation of ATF2 and the ensuing expression of PGC-1α and UCP1. Importantly, administration of IL-27 ameliorates obesity and insulin resistance in mice, and reduced serum IL-27 is correlated with the development of obesity in humans. Taken together, these findings provide new insights into IL-27 biology, expand our understanding of thermogenic control, and reveal IL-27 as a biologic agent with promising potential for anti-obesity immunotherapy.

Methods

Mice

Il27ra-KO (B6N.129P2-Il27ratm1Mak/J, 018078), Ebi3-KO (B6.129X1-Ebi3tm1Rsb/J, 008691), Ucp1-KO (B6.129-Ucp1tm1Kz/J, 003124), Cd2-cre (B6.Cg-Tg(CD2-icre)4Kio/J, 008520), Lyz2-cre (B6.129P2-Lyz2tm1(cre)Ifo/J, 004781), Itgax-cre (B6.Cg-Tg(Itgax-cre)1-1Reiz/J, 008068), Cx3cr1-cre (B6J.B6N(Cg)-Cx3cr1tm1.1(cre)Jung/J, 025524), C57BL/6J (000664) and Adipoq-cre (B6.FVB-Tg(Adipoq-cre)1Evdr/J, 028020) mice were purchased from The Jackson Laboratory. Ucp1-creERT2 mice22 were from C. Wolfrum (ETH Zurich) provided by X. Yang (Yale University). IL27raflox/flox (this paper) and Il27p28flox/flox mice23 were generated in our laboratory. Animal experiments were performed according to ethical regulations and protocols approved by the Institutional Animal Care and Use Committee of Jinan University and the Institutional Animal Care and Use Committee of Yale University. All of the experiments used sex- and age-matched mice that were group-housed at four to six animals per cage. The animal studies in this work have been carried out on either sex and yield consistent results. Randomization was performed in all animal experiments. No statistical methods were used to predetermine the sample size. Mice were group housed in a temperature- and humidity-controlled, specific-pathogen-free animal facility at 25 °C under a 12 h–12 h light–dark cycle with free access to food and water. For the diet study, mice (aged 8 weeks) were fed a 60% HFD (Research Diets, D12492) for the indicated times. Mouse body weights were measured every week.

Human samples

This study was approved by at the Institutional Review Board of the First Affiliated Hospital of Jinan University and was performed in accordance with the principle of the Helsinki Declaration II. Written informed consent was obtained from each participant. We recruited 42 individuals with obesity (BMI, 38.92 ± 1.349 kg m−2, mean ±  s.e.m., 42 for the Luminex immunoassay (Extended Data Fig. 1a) and 32 for intact IL-27 detection by ELISA (Fig. 1a and Extended Data Fig. 1b)), and 30 individuals with a healthy BMI (BMI, 20.69 ± 0.3144 kg m−2, 26 for Luminex immunoassay (Extended Data Fig. 1a) and 30 for intact IL-27 detection by ELISA (Fig. 1a)). Information regarding the characteristics of the human cohorts is provided in Supplementary Table 1. Seven of the individuals with obesity (BMI between 37 and 51) underwent Roux En Y Gastric Bypass surgery. Information regarding the characteristics of each patient before and after surgery are provided in Supplementary Table 3. Bariatric procedures were performed laparoscopically by one single surgeon. The participants who had undergone Roux En Y Gastric Bypass surgery were evaluated by surgical dietitian treatment. We also recruited 12 individuals who had been diagnosed with type 2 diabetes (Supplementary Table 2). The exclusion criteria for recruitment were: hypertension; abdominal surgery; previous bariatric surgery; virus hepatitis; colitis; gastrointestinal disease and gastrointestinal surgery within 5 years before recruitment; and abnormal liver and kidney function. All of the participants were weighed in light clothing without shoes. Body height and weight were measured by a height–weight scale, and BMI (kg m−2) was calculated.

Reagents

The human serum samples were analysed on the Bio-Plex system (BioRad) using a Bio-Plex Pro Human Cytokine 17-plex Assay (M5000031YV) and a Bio-Plex Pro Human Inflammation Panel 1 (171-AL001M). The Human IL-27 ELISA kit (434607) was purchased from BioLegend. Mouse serum samples were analysed by the Bio-Plex system using the 23-plex cytokine array kit (M60009RDPD). Glucose (63005518) was purchased from Sinopharm Chemical Reagent. Blood glucose strips (one-touch, accuracy to 0.1mg dl−1) were purchased from Johnson & Johnson. Anti-p-AKT (S473) monoclonal antibodies (4060, D9E), anti-mouse AKT monoclonal antibodies (4685, 11E7), anti-GAPDH monoclonal antibodies (5174, D16H11), anti-HSP90 monoclonal antibodies (4877, C45G5), anti-PPARγ monoclonal antibodies (2435, C26H12), anti-mouse p-STAT3 (Y705) monoclonal antibodies (9145, D3A7), anti-STAT3 monoclonal antibodies (9139, 124H6), anti-p-p38 MAPK (T180/Y182) polyclonal antibodies (9211), anti-p38 MAPK polyclonal antibodies (9212), anti-p-STAT1 (Y701) monoclonal antibodies (9167, 58D6), anti-STAT1 monoclonal antibodies (14994, D1K9Y), anti-pATF2 (T71) polyclonal antibodies (9221) and anti-ATF2 monoclonal antibodies (35031, D4L2X) were purchased from Cell Signaling Technology. Anti-UCP1 polyclonal antibodies (ab10983), anti-mouse UCP1 monoclonal antibodies (ab209483, EPR20381), anti-PGC-1α polyclonal antibodies (ab54481), anti-PPARα polyclonal antibodies (ab24509), anti-mouse IL-27Rα monoclonal antibodies (ab220359, EPR20863-3, 110 kDa) and anti-IL-27Rα polyclonal antibodies (ab5997, 70 kDa) were purchased from Abcam. Anti-β-actin monoclonal antibodies (66009-1, 2D4H5) and anti-β-tubulin monoclonal antibodies (66240-1, 1D4A4) were purchased from Proteintech. APC/CY7-anti-mouse CD45 (557659, 30-F11), BV510-anti-mouse CD45.2 (740131, 104) and PE/CY7-anti-mouse CD11b (552850, M1/70) antibodies were purchased from BD Biosciences. PE-anti-mouse F4/80 (M100F1-09B, BM8), PE-anti-mouse CD8α (M10081-09B, 53.6.7), PE-anti-mouse IL-17 (M100I17-09B, 17F3) and APC-anti-mouse IFN-γ (M100I16-11A, XMG1.2) antibodies were purchased from Sungene Biotech. PerCP/CY5.5-anti-mouse CD4 (100434, GK1.5), FITC-anti-mouse CD19 (152404, 1D3/CD19), APC/CY7-anti-mouse CD45.1 (110716, A20), FITC-anti-mouse CD8α (100706, 53.6.7), BV421-anti-mouse IL-10 (505022, JES5-16E3), PE/CY7-anti-mouse CD3 (100220, 17A2) and Alexa Fluor 647-anti-mouse F4/80 (123122, BM8) antibodies were purchased from BioLegend. Recombined murine IL-27 (rmIL-27) (577408) was purchased from BioLegend. STAT3 inhibitors C188-9, stattic and HO-3867, and p38 MAPK inhibitors SB203580 and SB202190 were purchased from Selleck Chemicals. BODIPY TR Ceramide (D7540) was purchased from Invitrogen. The brown/beige adipose differentiation agents dexamethasone (D4902), rosiglitazone (R2408), indomethacin (I7378), isoproterenol (I5627), forskolin (F6886), 3-isobutyl-1-methylxanthine (I5879), triglyceride detection kit (T2449) and insulin (I3536) were purchased from Sigma-Aldrich. The leptin detection kit (MOB00B) and adiponectin detection kit (MRP300) were purchased from R&D, and the insulin detection kit (90080) was purchased from Crystal Chem.

ELISA assays

Human serum was obtained as described above. Mouse serum was collected after feeding mice an HFD for 10 weeks. Assays using the ELISA kits for human IL-27 (434607, BioLegend), mouse leptin (MOB00B, R&D), mouse adiponectin (MRP300, R&D) and mouse insulin (90080, Crystal Chem) were performed according to the user manuals.

Luminex immunoassay

Human serum was collected as mentioned above. The inflammatory factors in human serum were detected by Bio-Plex Pro Human Cytokine 17-plex Assay (M5000031YV) and a Bio-Plex Pro Human Inflammation Panel 1 (171-AL001M) using Bio-Plex 200 Multiplexing Analyzer System (Bio-Rad) as instructed by the user manuals. Mouse serum samples were analysed by the Bio-Plex system using the 23-plex cytokine array kit (M60009RDPD) as instructed by the user manuals.

GTT

Mice were fasted for 12–18 h. Glucose (1 g kg−1) was administered intraperitoneally (i.p.), and the blood glucose levels were measured at 0 min, 30 min, 60 min, 90 min and 120 min using blood glucose strips (Johnson & Johnson, accuracy to 0.1 mg dl−1). The blood glucose level at 0 min was designated as the fasting glucose level.

ITT

Mice were fasted for 2–4 h and i.p. injected with insulin (0.5 U kg−1), blood glucose levels were measured at 0 min, 15 min, 30 min, 45 min, 60 min, 75 min, 90 min and 120 min using blood glucose strips. For the insulin sensitivity analysis, mice were fasted overnight, i.p. injected with insulin (0.75 U kg−1), and epididymal fat was isolated at 15 min after injection and used for immunoblotting analysis.

Triglyceride quantification

The concentration of triglycerides in the serum was quantified using a serum triglyceride determination kit (Sigma-Aldrich, Triglyceride Reagent, T2449, and Free Glycerol Reagent, F6428). For detection of triglycerides in the liver, 20–30 mg of tissue was homogenized in 500 μl of PBS and mixed with chloroform–methanol (2:1 (v/v)). The organic phase was transferred, air-dried overnight and resuspended in 1% Triton X-100 in absolute ethanol. The concentration of triglycerides was then quantified using the triglyceride determination kit.

Serum biochemistry

Mice were fasted overnight. Whole blood was next collected, and serum cholesterol levels were determined using an automatic biochemistry analyser (7600-020, Hitachi).

Western blot analysis

Whole-cell lysates or tissue lysates were extracted using RIPA lysis buffer (Beyotime) supplemented with complete protease inhibitor (Roche) and Phosphatase Inhibitor Cocktail 2 (P5726, Sigma-Aldrich), and the supernatants were used for the subsequent analysis. Proteins were diluted in loading dye (BL502A, Bio-sharp), heated at 95 °C for 10 min and run on 4–12% polyacrylamide gel. Proteins were transferred onto a polyvinylidene difluoride membrane and blotted with commercial antibodies mentioned in the figure legends and the ‘Reagents’ section. Each lane represents one biological independent sample in the immunoblot data. The band densities (grey values) for proteins of interest and housekeeping proteins were quantified using Image-Pro Plus v.6.0, the ratios of interested proteins and housekeeping proteins were calculated; for protein phosphorylation, the ratios of phosphorylated protein and total protein of interest were calculated, and the ratios were normalized and shown. The uncropped and unprocessed scans of immunoblot data are provided in Supplementary Fig. 1.

Histology

Adipose, liver or the indicated tissues were fixed in 4% paraformaldehyde/1× PBS overnight at 4 °C and embedded in paraffin before sectioning. Sections were stained with H&E or Oil Red O (liver) and photographed under bright-field microscopy. A representative image for each group was shown in our study.

Immunofluorescence

Cells were fixed in 4% paraformaldehyde for 10 min. After being blocked with 3% BSA/PBS, the cells were stained with an appropriate combination of primary antibodies, followed by staining with the corresponding fluorophore-conjugated secondary antibodies. The cells were mounted on coverslips and examined under a confocal microscope (Leica). Representative images for two independent experiments were shown in this work.

Immunohistochemistry

The rabbit-Specific HRP/DAB (ABC) detection IHC Kit (ab64261, Abcam) was used for UCP1 immunohistochemistry according to the manufacturer’s instructions. At least five biologically independent samples in each group were analysed and yielded similar results. A representative image for each group is shown.

Cold challenge and core body temperature measurement

Cold-challenge experiments were performed within climate-controlled cold rooms. Mice were singly placed in individual precooled cages without bedding at 4 °C. Mice had free access to precooled food and water. Rectal core body temperatures were recorded every 2 h using a digital thermometer and rectal thermocouple probe (TH-212, HICHANCE). Individual mice were euthanized if their core body temperature fell below 20 °C, and scored as an event for the survival analysis.

Metabolic cages

The energy consumption and energy expenditure of mice were measured using the Comprehensive Laboratory Animal Monitoring System (CLAMS, Columbus Instruments) metabolic cages housed within environment-controlled rodent incubators at Yale University. Mice were singly housed and acclimatized in metabolic chambers for 48 h before data collection. Mice had free access to food and water. Each mouse was continuously monitored for physical activity and food intake. CO2 and O2 levels were collected four times per hour per mouse over the duration of the experiment.

RNA isolation and gene expression analysis

Total RNA was extracted from frozen tissue using TRNzol Universal Reagent (Tiangen) and quantified using a Nanodrop 2000 ultraviolet–visible spectrophotometer (Thermo Fisher Scientific). cDNA was prepared using 1 µg total RNA by a reverse transcription PCR using the PrimeScript RT Reagent Kit (TAKARA). Quantitative PCR was performed on cDNA using the TB Green Premix EX Taq II kit (TAKARA) with the CFX96 Real-Time PCR Detection System (Bio-Rad). Fold changes in expression were calculated with the ▵▵Ct method using mouse HPRT as the endogenous control. A list of primer pair sequences is provided in Supplementary Table 5.

Bone marrow chimera

Il27ra-KO and WT mice (aged 10 weeks) were used as donor of bone marrow cells. Bone marrow was isolated from the femurs of hind legs. After red blood cell lysis and strained through cell strainer (40 µm), cell pellets were washed three times with sterile PBS, counted and stored on ice for later injection. Recipient Il27ra-KO and WT mice (aged 10 weeks) were lethally irradiated with 900 rads and transplanted with 1 × 107 bone marrow cells through ophthalmic vein injection. Mice were placed in specific-pathogen-free facilities supplemented with sterilized water and chow feed for 8 weeks to reconstitute the immune system and then processed for metabolism studies.

Primary beige adipocytes preparation

Inguinal subcutaneous adipose tissue was minced and digested for 45 min at 37 °C in PBS containing collagenase II (1 mg ml−1). Tissue suspension was filtered through a 100 µm cell strainer and centrifuged at 600g for 5 min to pellet the stromal vascular fraction (SVF). The pellet was further strained through 40 µm cell strainer and plated onto collagen-coated plates. After overnight incubation, the supernatant containing unadherent cells was removed. Preadipocytes were grown to confluence in DMEM with 10% FBS plus insulin (5 µg ml−1). Confluent cells were induced to differentiate with dexamethasone (1 µM), 3-isobutyl-1-methylxanthine (0.5 mM), insulin (5 µg ml−1), indomethacin (125 nM) and rosiglitazone (1 µM) for 2 d, followed by insulin (5 µg ml−1) and triiodothyronine (1 nM) alone for another 5 d. The purity of cultured primary adipocytes was assessed by staining of CD45 followed by confocal microscopy. On day 7, cells were pretreated with and without IL-27 (100 ng ml−1) for 12–24 h and then treated with isoproterenol (10 µM) or forskolin (10 µM) for 4–6 h. Cells were then lysed and used for immunoblotting analysis. For protein phosphorylation analysis, IL-27 (100 ng ml−1) was added on day 7 for 0–120 min. For signalling inhibition experiments, STAT3 inhibitors (C188-9, 10 µM; stattic, 10 µM; or HO-3867, 20 µM) or p38 MAPK inhibitor (SB203580, 10 µM; or SB202190, 5 µM) were added into the culture medium 0.5 h before and during IL-27 treatment.

Oxygen consumption measurements

Freshly isolated mouse inguinal SCW or BAT were rinsed with XF-DMEM (containing 25 mM HEPES) and cut into small pieces (~3 mg for BAT and ~4 mg for SCW). After extensive washing, one piece of tissue was placed in each well of a XF24 Islet Capture Microplate (Seahorse Bioscience) and covered with the islet capture screen that allows free perfusion while minimizing tissue movement. XF assay medium (500 µl) was added, and the samples were analysed in the XF24 Analyzer. These experiments were performed using 4–5 pieces per tissue per mouse, five individual mice per group. For SCW, the reagent concentrations were as follows: oligomycin (80 µM), FCCP (72 µM), and rotenone and antimycin A (240 µM and 120 µM, respectively). For BAT, the reagent concentrations were as follows: oligomycin (160 µM), FCCP (240 µM), and rotenone and antimycin A (480 µM and 360 µM, respectively).

Flow cytometry analysis of immune cells

eWAT was minced and digested as described above. SVF pellets were used for surface staining of APC-CY7-anti mouse CD45, PE-Cy7-anti mouse CD11b and PE-anti mouse F4/80 antibodies. After incubation for 15 min, cells were washed with PBS and analysed using the BD FACSVerse Flow Cytometer (BD). The peripheral blood, inguinal lymph nodes, spleen, liver, BAT and SCW from bone marrow chimeric mice were collected and used for isolation of immune cells or SVF pellets. Cells were stained with APC-CY7-anti-mouse CD45.1, BV510-anti-mouse CD45.2, FITC-anti-mouse CD19, PerCP/CY5.5-anti-mouse CD4, PE-anti-mouse CD8α and Alexa Fluor 647-anti-mouse F4/80 antibodies. After incubation for 15 min, cells were washed with PBS and analysed using the BD FACSVerse Flow Cytometer (BD). FACS data were analysed using FlowJo v.10. The gating strategies are provided in Supplementary Fig. 2.

Generation of IL-27Rα conditional KO mice

The generation of Il27raflox/flox mice was performed by the Cyagen Biotechnology using gene-targeting technology. In brief, a 997 bp conditional KO region containing exon 3 and exon 4 of IL27ra gene (GenBank: NM_016671.3), a 5.2 kb 5′ homology arm and a 3 kb 3′ homology arm were amplified from a bacterial artificial chromosome clone using high fidelity Taq, and were sequentially assembled into a targeting vector together with recombination sites and a neomycin selection cassette. The final targeting vector was confirmed by digestion with multiple restriction enzymes and full sequencing, then transformed into C57BL/6 mice embryonic stem cells by electroporation. Correct embryonic stem cell clones were identified and injected into C57BL/6 mouse blastocysts. Chimaera mice were visibly identified and crossed with Flp-deleter to delete the neomycin selection cassette. The germ line transmission was confirmed by genotyping PCR. The offspring with the Il27raflox/WT genotype were then crossed with Cd2-cre, Lyz2-cre, Adipoq-cre or Ucp1-creERT2 mice. The mice with the expected genotype were used for experiment. The genotyping primers to detect the flox (331 bp) or WT (257 bp) band were as follows: forward: 5′-CTGGTTCTGGTATGGTTTGGGGTT-3′, reverse: 5′-TGAA AGAACTCAACAGTGGGCCGG-3′.

IL-27 administration

WT, Il27raflox/flox, Adipoq-cre;Il27raflox/flox, Ucp1-creERT2;Il27raflox/flox or Ucp1-KO mice (aged 8 weeks) were fed an HFD for 12–32 weeks to induce severe obesity. Ucp1-creERT2;Il27raflox/flox mice were pretreated with tamoxifen before IL-27 therapy (see below). Mice of each genotype were randomly divided into two groups for i.p. injection of rmIL-27 (100 µg kg−1) or PBS every other day for 15 d, and then processed for metabolic analysis. Ebi3-KO mice (aged 12 months) were randomly divided into two groups and i.p. injected with rmIL-27 (100 µg kg−1) or PBS every day for 7 d, and then placed at 4 °C for cold-challenge experiments.

Tamoxifen preparation and treatment

Tamoxifen (Sigma-Aldrich, T5468) was dissolved in sunflower seed oil/ethanol (9:1 v/v) at 20 mg ml−1 by shaking at 37 °C for 15 min. Mice (aged 8–10 weeks) were i.p. injected with 2 mg per mouse four times over a period of 7 d. After another 7 d, mice were subjected to HFD feeding, cold stimulation or IL-27 administration for the indicated time.

RNA sequencing

Total RNA was extracted using the mirVana miRNA Isolation Kit (Ambion) according to the manufacturer’s protocol. RNA integrity was evaluated using the Agilent 2100 Bioanalyzer (Agilent Technologies). The samples with an RNA integrity number of ≥7 were processed for the subsequent analysis. The transcriptome sequencing and analysis were conducted by OE biotech. Raw data (raw reads) were processed using Trimmomatic. The reads containing ploy-N and the low-quality reads were removed to obtain the clean reads. The clean reads were next mapped to the reference genome using hisat2. Fragments per kb of transcript per million mapped reads and read count values of each transcript (protein_coding) was calculated using bowtie2 and eXpress. Differentially expressed genes (DEGs) were identified using the DESeq (2012) functions estimatSizeFactors and nbinomTest. P < 0.05 and a fold change of >2 or <0.5 was set as the threshold for a significant differential expression pattern. Hierarchical cluster analysis of DEGs was performed to examine transcript expression patterns. Gene Ontology enrichment analysis and KEGG pathway enrichment analysis of DEGs were performed using R based on the hypergeometric distribution.

Statistics and reproducibility

All the non-human experiments and data shown in this work have been repeated at least twice with consistent results. The presented data were performed on biologically independent samples. GraphPad Prism (v.8) was used for graphing and statistical analysis. All statistical tests are fully described in the figure legends and met the criteria for normal distribution with similar variance. No statistical methods were used to predetermine sample sizes. Two tailed student’s t-tests were used for comparisons between two groups. For assessment between more than two groups, one-way ANOVA was used. For assessment between two independent variables, two-way ANOVA was used; Sidak’s multiple comparisons test was also performed when comparing the difference at each time point. For survival analysis, a log-rank test was performed. For correlation analysis, linear regression analysis was performed. Data are presented as mean (average) ± s.e.m. unless otherwise stated. P < 0.05 was considered to be statistically significant and the exact P values are provided in the figures.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this paper.